This paper presents a real-time multi-object tracking system by integrating the YOLOv10 object detection model with the ByteTrack data association algorithm. In the proposed system, YOLOv10 provides accurate object detections for each video frame, while ByteTrack performs data association using a two-stage matching strategy that exploits both high-confidence and low-confidence detections to reduce ID switches. Experimental results show that the system achieves a detection precision of 0,90, a recall of 0,85, an overall tracking accuracy of 0,82, and only 10 ID switches. When processing a 720p video, the system maintains a processing speed of 7-8 frames per second. These results indicate that the combination of YOLOv10 and ByteTrack demonstrates suitable performance for applications such as surveillance, intelligent transportation systems, and behavior analysis, and can be further improved through parameter optimization or the integration of lightweight re-identification modules.